Nonintrusive autonomous water monitoring System and water consumption disaggregation

Water monitoring is rather limited as it is in Belgium with the water meters being, for the vast majority, mechanical. Rationalization of resources is an actual concern and water is part of it. The objective of this master thesis is to address this problem by analyzing the signal generated by the mechanical water meter. This signal is recorded by the Sense Hat equipped on a Raspberry Pi. After determining a method to efficiently analyze the signal wave to get its frequency, it will be possible to get an idea of how much water is being used. Patterns (tap water and WC) will be identified with the k-nearest neighbors algorithm, which could allow the user to know the distribution of its water usage. Due to each patterns being really similar, the success rate of the pattern matching algorithm is good but not perfect.